A Common Variant at the 14q32 Endometrial Cancer Risk Locus Activates AKT1 through YY1 Binding

Abstract

A recent meta-analysis of multiple genome-wide association and follow-up endometrial cancer case-control datasets identified a novel genetic risk locus for this disease at chromosome 14q32.33. To prioritize the functional SNP(s) and target gene(s) at this locus, we employed an in silico fine-mapping approach using genotyped and imputed SNP data for 6,608 endometrial cancer cases and 37,925 controls of European ancestry. Association and functional analyses provide evidence that the best candidate causal SNP is rs2494737. Multiple experimental analyses show that SNP rs2494737 maps to a silencer element located within AKT1, a member of the PI3K/AKT/MTOR intracellular signaling pathway activated in endometrial tumors. The rs2494737 risk A allele creates a YY1 transcription factor-binding site and abrogates the silencer activity in luciferase assays, an effect mimicked by transfection of YY1 siRNA. Our findings suggest YY1 is a positive regulator of AKT1, mediating the stimulatory effects of rs2494737 increasing endometrial cancer risk. Identification of an endometrial cancer risk allele within a member of the PI3K/AKT signaling pathway, more commonly activated in tumors by somatic alterations, raises the possibility that well tolerated inhibitors targeting this pathway could be candidates for evaluation as chemopreventive agents in individuals at high risk of developing endometrial cancer.

Figure 1

Regional Association Plot for the 14q32.33 Endometrial Cancer Risk Locus The location (Build 19) and –log₁₀ p value of the original top SNP at this locus, rs2498796, is shown in purple; all other SNPs are shown in colors corresponding to their r² (linkage disequilibrium) value with rs2498796.

Figure 2

Regulatory Landscape at the 14q32.33 Endometrial Cancer Risk Locus (A) The location of the candidate SNPs are represented by black ticks, and the PRE is shown as a black box. Gene structures are depicted with exons (vertical boxes) joined by introns (lines). The subset of enhancers predicted in Hnisz et al. and PreSTIGE which overlap the candidate causal SNPs are shown as colored bars, where the color matches its predicted gene target. Regions showing histone binding (H3K4Me1, indicative of regulatory regions; H3K4Me3, indicative of promoters; and H3K27Ac, indicative of active enhancers), DNAseI hypersensitivity (indicative of open chromatin, with darker shading indicating stronger experimental signal) and transcription factor (TF) binding in multiple ENCODE cell lines are indicated at the bottom of the panel. (B) Zoomed-in view of the location of candidate SNPs, PRE, and nearby gene promoter regions.

Figure 3

Candidate Causal SNPs Are Located within a PRE that Interacts with the AKT1 and ZBTB42 Promoter Regions (A) 3C interaction profiles between the PRE and local genes in Ishikawa and EN-1078D endometrial cancer cell lines. The 3C anchor (which contains the PRE) is shown as a grey box and significant interactions are outlined. (B) Zoomed-in view of significant interactions. AKT1 Can and AKT1 Alt denote a canonical and alternative AKT1 promoter (prom) region, respectively. 3C libraries were generated with NcoI. Graphs represent three biological replicates. Error bars represent SD. Regions cloned into reporter gene constructs are shown as grey horizontal boxes.

Figure 4

The Risk Allele of SNP rs2494737 Enhances AKT1 Promoter Activity Luciferase reporter assays following transient transfection of Ishikawa endometrial cancer cell lines. The putative regulatory element (PRE) containing the major SNP alleles were cloned downstream of target gene promoter-driven luciferase constructs. AKT1 can and AKT1 alt denote a canonical and alternative AKT1 promoter (prom) region, respectively. Minor (risk-increasing) SNP alleles were engineered into the constructs and are designated by the rs ID of the corresponding SNP. Haplotype denotes a construct that contains the minor alleles of rs2498796 and rs2494737. Error bars denote 95% confidence intervals from three independent experiments performed in duplicate. P values were determined by 2-way ANOVA followed by Dunnett’s multiple comparisons test (^∗∗p < 0.01, ^∗∗∗∗p < 0.0001).

Figure 5

The Risk Allele of rs2494737 Demonstrates Allele-Specific YY1 Binding (A) EMSAs to detect allele-specific binding of nuclear proteins. Oligonucleotides were incubated with Ishikawa nuclear extracts. Red arrowheads show bands of different mobility detected between the common (C) and minor (risk-increasing) (m) alleles for the candidate causal SNPs. (B) Oligonucleotides for SNP rs2494737 were incubated with Ishikawa nuclear extracts. Red arrowhead indicates the band that was competed for complex formation on the minor (m) allele. Competitor oligonucleotides are listed above each panel and were used at 100-fold molar excess: (−) no competitor; (Neg) a non-specific competitor; (YY1) consensus binding site; (YY1m) an identical oligonucleotide but with a mutated binding site. (C) ChIP–qPCR on SNP rs2494737 in heterozygous Ishikawa cell lines. ChIP assays were performed with YY1 antibody or non-immune IgG, a region 3.2kb upstream of the predicted YY1-binding site served as a negative control (Control). Graphs represent two biological replicates. Error bars denote SD. P-values were determined with a two-tailed t-test (^∗∗p < 0.01). (D) Sanger sequencing of the PCR fragment generated using primers flanking SNP rs2494737 following YY1 ChIP-qPCR and the input DNA controls. (E) Luciferase assays in Ishikawa cells shows the effect of YY1 siRNA silencing on the activity of the AKT1 canonical (can) and alternative (alt) promoter regions with the PRE containing the reference T allele (can+PRE; alt+PRE) or the risk A allele (rs2494737). Error bars denote 95% confidence intervals from three independent experiments performed in duplicate. P values were determined by two-way ANOVA followed by Dunnett’s multiple comparisons test (^∗∗p < 0.01, ^∗∗∗p < 0.001). The level of YY1 silencing is shown in Figure S8.